DUK_LOCAL duk_ret_t duk__error_setter_helper(duk_context *ctx, duk_small_uint_t stridx_key) {
/* Attempt to write 'stack', 'fileName', 'lineNumber' works as if
* user code called Object.defineProperty() to create an overriding
* own property. This allows user code to overwrite .fileName etc
* intuitively as e.g. "err.fileName = 'dummy'" as one might expect.
* See https://github.com/svaarala/duktape/issues/387.
*/
DUK_ASSERT_TOP(ctx, 1); /* fixed arg count: value */
duk_push_this(ctx);
duk_push_hstring_stridx(ctx, (duk_small_int_t) stridx_key);
duk_dup_0(ctx);
/* [ ... obj key value ] */
DUK_DD(DUK_DDPRINT("error setter: %!T %!T %!T",
duk_get_tval(ctx, -3), duk_get_tval(ctx, -2), duk_get_tval(ctx, -1)));
duk_def_prop(ctx, -3, DUK_DEFPROP_HAVE_VALUE |
DUK_DEFPROP_HAVE_WRITABLE | DUK_DEFPROP_WRITABLE |
DUK_DEFPROP_HAVE_ENUMERABLE | /*not enumerable*/
DUK_DEFPROP_HAVE_CONFIGURABLE | DUK_DEFPROP_CONFIGURABLE);
return 0;
}
DUK_INTERNAL duk_ret_t duk_bi_string_constructor(duk_context *ctx) {
/* String constructor needs to distinguish between an argument not given at all
* vs. given as 'undefined'. We're a vararg function to handle this properly.
*/
if (duk_get_top(ctx) == 0) {
duk_push_hstring_stridx(ctx, DUK_STRIDX_EMPTY_STRING);
} else {
duk_to_string(ctx, 0);
}
DUK_ASSERT(duk_is_string(ctx, 0));
duk_set_top(ctx, 1);
if (duk_is_constructor_call(ctx)) {
duk_push_object_helper(ctx,
DUK_HOBJECT_FLAG_EXTENSIBLE |
DUK_HOBJECT_FLAG_EXOTIC_STRINGOBJ |
DUK_HOBJECT_CLASS_AS_FLAGS(DUK_HOBJECT_CLASS_STRING),
DUK_BIDX_STRING_PROTOTYPE);
/* String object internal value is immutable */
duk_dup_0(ctx);
duk_xdef_prop_stridx(ctx, -2, DUK_STRIDX_INT_VALUE, DUK_PROPDESC_FLAGS_NONE);
}
/* Note: unbalanced stack on purpose */
return 1;
}
DUK_INTERNAL duk_ret_t duk_bi_object_prototype_to_locale_string(duk_hthread *thr) {
DUK_ASSERT_TOP(thr, 0);
(void) duk_push_this_coercible_to_object(thr);
duk_get_prop_stridx_short(thr, 0, DUK_STRIDX_TO_STRING);
#if 0 /* This is mentioned explicitly in the E5.1 spec, but duk_call_method() checks for it in practice. */
duk_require_callable(thr, 1);
#endif
duk_dup_0(thr); /* -> [ O toString O ] */
duk_call_method(thr, 0); /* XXX: call method tail call? */
return 1;
}
DUK_INTERNAL duk_ret_t duk_bi_error_constructor_shared(duk_context *ctx) {
/* Behavior for constructor and non-constructor call is
* the same except for augmenting the created error. When
* called as a constructor, the caller (duk_new()) will handle
* augmentation; when called as normal function, we need to do
* it here.
*/
duk_hthread *thr = (duk_hthread *) ctx;
duk_small_int_t bidx_prototype = duk_get_current_magic(ctx);
/* same for both error and each subclass like TypeError */
duk_uint_t flags_and_class = DUK_HOBJECT_FLAG_EXTENSIBLE |
DUK_HOBJECT_CLASS_AS_FLAGS(DUK_HOBJECT_CLASS_ERROR);
DUK_UNREF(thr);
duk_push_object_helper(ctx, flags_and_class, bidx_prototype);
/* If message is undefined, the own property 'message' is not set at
* all to save property space. An empty message is inherited anyway.
*/
if (!duk_is_undefined(ctx, 0)) {
duk_to_string(ctx, 0);
duk_dup_0(ctx); /* [ message error message ] */
duk_xdef_prop_stridx(ctx, -2, DUK_STRIDX_MESSAGE, DUK_PROPDESC_FLAGS_WC);
}
/* Augment the error if called as a normal function. __FILE__ and __LINE__
* are not desirable in this case.
*/
#ifdef DUK_USE_AUGMENT_ERROR_CREATE
if (!duk_is_constructor_call(ctx)) {
duk_err_augment_error_create(thr, thr, NULL, 0, 1 /*noblame_fileline*/);
}
#endif
return 1;
}
/* XXX: much to improve (code size) */
DUK_INTERNAL duk_ret_t duk_bi_regexp_constructor(duk_hthread *thr) {
duk_hobject *h_pattern;
DUK_ASSERT_TOP(thr, 2);
h_pattern = duk_get_hobject(thr, 0);
if (!duk_is_constructor_call(thr) &&
h_pattern != NULL &&
DUK_HOBJECT_GET_CLASS_NUMBER(h_pattern) == DUK_HOBJECT_CLASS_REGEXP &&
duk_is_undefined(thr, 1)) {
/* Called as a function, pattern has [[Class]] "RegExp" and
* flags is undefined -> return object as is.
*/
/* XXX: ES2015 has a NewTarget SameValue() check which is not
* yet implemented.
*/
duk_dup_0(thr);
return 1;
}
/* Else functionality is identical for function call and constructor
* call.
*/
if (h_pattern != NULL &&
DUK_HOBJECT_GET_CLASS_NUMBER(h_pattern) == DUK_HOBJECT_CLASS_REGEXP) {
duk_get_prop_stridx_short(thr, 0, DUK_STRIDX_SOURCE);
if (duk_is_undefined(thr, 1)) {
/* In ES5 one would need to read the flags individually;
* in ES2015 just read .flags.
*/
duk_get_prop_stridx(thr, 0, DUK_STRIDX_FLAGS);
} else {
/* In ES2015 allowed; overrides argument RegExp flags. */
duk_dup_1(thr);
}
} else {
if (duk_is_undefined(thr, 0)) {
duk_push_hstring_empty(thr);
} else {
duk_dup_0(thr);
duk_to_string(thr, -1); /* Rejects Symbols. */
}
if (duk_is_undefined(thr, 1)) {
duk_push_hstring_empty(thr);
} else {
duk_dup_1(thr);
duk_to_string(thr, -1); /* Rejects Symbols. */
}
/* [ ... pattern flags ] */
}
DUK_DDD(DUK_DDDPRINT("RegExp constructor/function call, pattern=%!T, flags=%!T",
(duk_tval *) duk_get_tval(thr, -2), (duk_tval *) duk_get_tval(thr, -1)));
/* [ ... pattern flags ] (both uncoerced) */
duk_to_string(thr, -2);
duk_to_string(thr, -1);
duk_regexp_compile(thr);
/* [ ... bytecode escaped_source ] */
duk_regexp_create_instance(thr);
/* [ ... RegExp ] */
return 1;
}
DUK_INTERNAL duk_ret_t duk_bi_object_constructor_define_properties(duk_hthread *thr) {
duk_small_uint_t pass;
duk_uint_t defprop_flags;
duk_hobject *obj;
duk_idx_t idx_value;
duk_hobject *get;
duk_hobject *set;
/* Lightfunc and plain buffer handling by ToObject() coercion. */
obj = duk_require_hobject_promote_mask(thr, 0, DUK_TYPE_MASK_LIGHTFUNC | DUK_TYPE_MASK_BUFFER);
DUK_ASSERT(obj != NULL);
duk_to_object(thr, 1); /* properties object */
DUK_DDD(DUK_DDDPRINT("target=%!iT, properties=%!iT",
(duk_tval *) duk_get_tval(thr, 0),
(duk_tval *) duk_get_tval(thr, 1)));
/*
* Two pass approach to processing the property descriptors.
* On first pass validate and normalize all descriptors before
* any changes are made to the target object. On second pass
* make the actual modifications to the target object.
*
* Right now we'll just use the same normalize/validate helper
* on both passes, ignoring its outputs on the first pass.
*/
for (pass = 0; pass < 2; pass++) {
duk_set_top(thr, 2); /* -> [ hobject props ] */
duk_enum(thr, 1, DUK_ENUM_OWN_PROPERTIES_ONLY | DUK_ENUM_INCLUDE_SYMBOLS /*enum_flags*/);
for (;;) {
duk_hstring *key;
/* [ hobject props enum(props) ] */
duk_set_top(thr, 3);
if (!duk_next(thr, 2, 1 /*get_value*/)) {
break;
}
DUK_DDD(DUK_DDDPRINT("-> key=%!iT, desc=%!iT",
(duk_tval *) duk_get_tval(thr, -2),
(duk_tval *) duk_get_tval(thr, -1)));
/* [ hobject props enum(props) key desc ] */
duk_hobject_prepare_property_descriptor(thr,
4 /*idx_desc*/,
&defprop_flags,
&idx_value,
&get,
&set);
/* [ hobject props enum(props) key desc [multiple values] ] */
if (pass == 0) {
continue;
}
/* This allows symbols on purpose. */
key = duk_known_hstring(thr, 3);
DUK_ASSERT(key != NULL);
duk_hobject_define_property_helper(thr,
defprop_flags,
obj,
key,
idx_value,
get,
set,
1 /*throw_flag*/);
}
}
/*
* Return target object
*/
duk_dup_0(thr);
return 1;
}
DUK_INTERNAL duk_ret_t duk_bi_string_prototype_replace(duk_context *ctx) {
duk_hthread *thr = (duk_hthread *) ctx;
duk_hstring *h_input;
duk_hstring *h_match;
duk_hstring *h_search;
duk_hobject *h_re;
duk_bufwriter_ctx bw_alloc;
duk_bufwriter_ctx *bw;
#ifdef DUK_USE_REGEXP_SUPPORT
duk_bool_t is_regexp;
duk_bool_t is_global;
#endif
duk_bool_t is_repl_func;
duk_uint32_t match_start_coff, match_start_boff;
#ifdef DUK_USE_REGEXP_SUPPORT
duk_int_t match_caps;
#endif
duk_uint32_t prev_match_end_boff;
const duk_uint8_t *r_start, *r_end, *r; /* repl string scan */
duk_size_t tmp_sz;
DUK_ASSERT_TOP(ctx, 2);
h_input = duk_push_this_coercible_to_string(ctx);
DUK_ASSERT(h_input != NULL);
bw = &bw_alloc;
DUK_BW_INIT_PUSHBUF(thr, bw, DUK_HSTRING_GET_BYTELEN(h_input)); /* input size is good output starting point */
DUK_ASSERT_TOP(ctx, 4);
/* stack[0] = search value
* stack[1] = replace value
* stack[2] = input string
* stack[3] = result buffer
*/
h_re = duk_get_hobject_with_class(ctx, 0, DUK_HOBJECT_CLASS_REGEXP);
if (h_re) {
#ifdef DUK_USE_REGEXP_SUPPORT
is_regexp = 1;
is_global = duk_get_prop_stridx_boolean(ctx, 0, DUK_STRIDX_GLOBAL, NULL);
if (is_global) {
/* start match from beginning */
duk_push_int(ctx, 0);
duk_put_prop_stridx(ctx, 0, DUK_STRIDX_LAST_INDEX);
}
#else /* DUK_USE_REGEXP_SUPPORT */
DUK_DCERROR_UNSUPPORTED(thr);
#endif /* DUK_USE_REGEXP_SUPPORT */
} else {
duk_to_string(ctx, 0);
#ifdef DUK_USE_REGEXP_SUPPORT
is_regexp = 0;
is_global = 0;
#endif
}
if (duk_is_function(ctx, 1)) {
is_repl_func = 1;
r_start = NULL;
r_end = NULL;
} else {
duk_hstring *h_repl;
is_repl_func = 0;
h_repl = duk_to_hstring(ctx, 1);
DUK_ASSERT(h_repl != NULL);
r_start = DUK_HSTRING_GET_DATA(h_repl);
r_end = r_start + DUK_HSTRING_GET_BYTELEN(h_repl);
}
prev_match_end_boff = 0;
for (;;) {
/*
* If matching with a regexp:
* - non-global RegExp: lastIndex not touched on a match, zeroed
* on a non-match
* - global RegExp: on match, lastIndex will be updated by regexp
* executor to point to next char after the matching part (so that
* characters in the matching part are not matched again)
*
* If matching with a string:
* - always non-global match, find first occurrence
*
* We need:
* - The character offset of start-of-match for the replacer function
* - The byte offsets for start-of-match and end-of-match to implement
* the replacement values $&, $`, and $', and to copy non-matching
* input string portions (including header and trailer) verbatim.
*
* NOTE: the E5.1 specification is a bit vague how the RegExp should
* behave in the replacement process; e.g. is matching done first for
* all matches (in the global RegExp case) before any replacer calls
* are made? See: test-bi-string-proto-replace.js for discussion.
*/
DUK_ASSERT_TOP(ctx, 4);
#ifdef DUK_USE_REGEXP_SUPPORT
if (is_regexp) {
duk_dup_0(ctx);
duk_dup_2(ctx);
duk_regexp_match(thr); /* [ ... regexp input ] -> [ res_obj ] */
if (!duk_is_object(ctx, -1)) {
duk_pop(ctx);
break;
}
duk_get_prop_stridx(ctx, -1, DUK_STRIDX_INDEX);
DUK_ASSERT(duk_is_number(ctx, -1));
match_start_coff = duk_get_int(ctx, -1);
duk_pop(ctx);
duk_get_prop_index(ctx, -1, 0);
DUK_ASSERT(duk_is_string(ctx, -1));
h_match = duk_known_hstring(ctx, -1);
duk_pop(ctx); /* h_match is borrowed, remains reachable through match_obj */
if (DUK_HSTRING_GET_BYTELEN(h_match) == 0) {
/* This should be equivalent to match() algorithm step 8.f.iii.2:
* detect an empty match and allow it, but don't allow it twice.
*/
duk_uint32_t last_index;
duk_get_prop_stridx(ctx, 0, DUK_STRIDX_LAST_INDEX);
last_index = (duk_uint32_t) duk_get_uint(ctx, -1);
DUK_DDD(DUK_DDDPRINT("empty match, bump lastIndex: %ld -> %ld",
(long) last_index, (long) (last_index + 1)));
duk_pop(ctx);
duk_push_int(ctx, last_index + 1);
duk_put_prop_stridx(ctx, 0, DUK_STRIDX_LAST_INDEX);
}
DUK_ASSERT(duk_get_length(ctx, -1) <= DUK_INT_MAX); /* string limits */
match_caps = (duk_int_t) duk_get_length(ctx, -1);
} else {
#else /* DUK_USE_REGEXP_SUPPORT */
{ /* unconditionally */
#endif /* DUK_USE_REGEXP_SUPPORT */
const duk_uint8_t *p_start, *p_end, *p; /* input string scan */
const duk_uint8_t *q_start; /* match string */
duk_size_t q_blen;
#ifdef DUK_USE_REGEXP_SUPPORT
DUK_ASSERT(!is_global); /* single match always */
#endif
p_start = DUK_HSTRING_GET_DATA(h_input);
p_end = p_start + DUK_HSTRING_GET_BYTELEN(h_input);
p = p_start;
h_search = duk_known_hstring(ctx, 0);
q_start = DUK_HSTRING_GET_DATA(h_search);
q_blen = (duk_size_t) DUK_HSTRING_GET_BYTELEN(h_search);
p_end -= q_blen; /* ensure full memcmp() fits in while */
match_start_coff = 0;
while (p <= p_end) {
DUK_ASSERT(p + q_blen <= DUK_HSTRING_GET_DATA(h_input) + DUK_HSTRING_GET_BYTELEN(h_input));
if (DUK_MEMCMP((const void *) p, (const void *) q_start, (size_t) q_blen) == 0) {
duk_dup_0(ctx);
h_match = duk_known_hstring(ctx, -1);
#ifdef DUK_USE_REGEXP_SUPPORT
match_caps = 0;
#endif
goto found;
}
/* track utf-8 non-continuation bytes */
if ((p[0] & 0xc0) != 0x80) {
match_start_coff++;
}
p++;
}
/* not found */
break;
}
found:
/* stack[0] = search value
* stack[1] = replace value
* stack[2] = input string
* stack[3] = result buffer
* stack[4] = regexp match OR match string
*/
match_start_boff = duk_heap_strcache_offset_char2byte(thr, h_input, match_start_coff);
tmp_sz = (duk_size_t) (match_start_boff - prev_match_end_boff);
DUK_BW_WRITE_ENSURE_BYTES(thr, bw, DUK_HSTRING_GET_DATA(h_input) + prev_match_end_boff, tmp_sz);
prev_match_end_boff = match_start_boff + DUK_HSTRING_GET_BYTELEN(h_match);
if (is_repl_func) {
duk_idx_t idx_args;
duk_hstring *h_repl;
/* regexp res_obj is at index 4 */
duk_dup_1(ctx);
idx_args = duk_get_top(ctx);
#ifdef DUK_USE_REGEXP_SUPPORT
if (is_regexp) {
duk_int_t idx;
duk_require_stack(ctx, match_caps + 2);
for (idx = 0; idx < match_caps; idx++) {
/* match followed by capture(s) */
duk_get_prop_index(ctx, 4, idx);
}
} else {
#else /* DUK_USE_REGEXP_SUPPORT */
{ /* unconditionally */
#endif /* DUK_USE_REGEXP_SUPPORT */
/* match == search string, by definition */
duk_dup_0(ctx);
}
duk_push_int(ctx, match_start_coff);
duk_dup_2(ctx);
/* [ ... replacer match [captures] match_char_offset input ] */
duk_call(ctx, duk_get_top(ctx) - idx_args);
h_repl = duk_to_hstring(ctx, -1); /* -> [ ... repl_value ] */
DUK_ASSERT(h_repl != NULL);
DUK_BW_WRITE_ENSURE_HSTRING(thr, bw, h_repl);
duk_pop(ctx); /* repl_value */
} else {
r = r_start;
while (r < r_end) {
duk_int_t ch1;
duk_int_t ch2;
#ifdef DUK_USE_REGEXP_SUPPORT
duk_int_t ch3;
#endif
duk_size_t left;
ch1 = *r++;
if (ch1 != DUK_ASC_DOLLAR) {
goto repl_write;
}
left = r_end - r;
if (left <= 0) {
goto repl_write;
}
ch2 = r[0];
switch (ch2) {
case DUK_ASC_DOLLAR: {
ch1 = (1 << 8) + DUK_ASC_DOLLAR;
goto repl_write;
}
case DUK_ASC_AMP: {
DUK_BW_WRITE_ENSURE_HSTRING(thr, bw, h_match);
r++;
continue;
}
case DUK_ASC_GRAVE: {
tmp_sz = (duk_size_t) match_start_boff;
DUK_BW_WRITE_ENSURE_BYTES(thr, bw, DUK_HSTRING_GET_DATA(h_input), tmp_sz);
r++;
continue;
}
case DUK_ASC_SINGLEQUOTE: {
duk_uint32_t match_end_boff;
/* Use match charlen instead of bytelen, just in case the input and
* match codepoint encodings would have different lengths.
*/
match_end_boff = duk_heap_strcache_offset_char2byte(thr,
h_input,
match_start_coff + DUK_HSTRING_GET_CHARLEN(h_match));
tmp_sz = (duk_size_t) (DUK_HSTRING_GET_BYTELEN(h_input) - match_end_boff);
DUK_BW_WRITE_ENSURE_BYTES(thr, bw, DUK_HSTRING_GET_DATA(h_input) + match_end_boff, tmp_sz);
r++;
continue;
}
default: {
#ifdef DUK_USE_REGEXP_SUPPORT
duk_int_t capnum, captmp, capadv;
/* XXX: optional check, match_caps is zero if no regexp,
* so dollar will be interpreted literally anyway.
*/
if (!is_regexp) {
goto repl_write;
}
if (!(ch2 >= DUK_ASC_0 && ch2 <= DUK_ASC_9)) {
goto repl_write;
}
capnum = ch2 - DUK_ASC_0;
capadv = 1;
if (left >= 2) {
ch3 = r[1];
if (ch3 >= DUK_ASC_0 && ch3 <= DUK_ASC_9) {
captmp = capnum * 10 + (ch3 - DUK_ASC_0);
if (captmp < match_caps) {
capnum = captmp;
capadv = 2;
}
}
}
if (capnum > 0 && capnum < match_caps) {
DUK_ASSERT(is_regexp != 0); /* match_caps == 0 without regexps */
/* regexp res_obj is at offset 4 */
duk_get_prop_index(ctx, 4, (duk_uarridx_t) capnum);
if (duk_is_string(ctx, -1)) {
duk_hstring *h_tmp_str;
h_tmp_str = duk_known_hstring(ctx, -1);
DUK_BW_WRITE_ENSURE_HSTRING(thr, bw, h_tmp_str);
} else {
/* undefined -> skip (replaced with empty) */
}
duk_pop(ctx);
r += capadv;
continue;
} else {
goto repl_write;
}
#else /* DUK_USE_REGEXP_SUPPORT */
goto repl_write; /* unconditionally */
#endif /* DUK_USE_REGEXP_SUPPORT */
} /* default case */
} /* switch (ch2) */
repl_write:
/* ch1 = (r_increment << 8) + byte */
DUK_BW_WRITE_ENSURE_U8(thr, bw, (duk_uint8_t) (ch1 & 0xff));
r += ch1 >> 8;
} /* while repl */
} /* if (is_repl_func) */
duk_pop(ctx); /* pop regexp res_obj or match string */
#ifdef DUK_USE_REGEXP_SUPPORT
if (!is_global) {
#else
{ /* unconditionally; is_global==0 */
#endif
break;
}
}
/* trailer */
tmp_sz = (duk_size_t) (DUK_HSTRING_GET_BYTELEN(h_input) - prev_match_end_boff);
DUK_BW_WRITE_ENSURE_BYTES(thr, bw, DUK_HSTRING_GET_DATA(h_input) + prev_match_end_boff, tmp_sz);
DUK_ASSERT_TOP(ctx, 4);
DUK_BW_COMPACT(thr, bw);
(void) duk_buffer_to_string(ctx, -1);
return 1;
}
/*
* split()
*/
/* XXX: very messy now, but works; clean up, remove unused variables (nomimally
* used so compiler doesn't complain).
*/
DUK_INTERNAL duk_ret_t duk_bi_string_prototype_split(duk_context *ctx) {
duk_hthread *thr = (duk_hthread *) ctx;
duk_hstring *h_input;
duk_hstring *h_sep;
duk_uint32_t limit;
duk_uint32_t arr_idx;
#ifdef DUK_USE_REGEXP_SUPPORT
duk_bool_t is_regexp;
#endif
duk_bool_t matched; /* set to 1 if any match exists (needed for empty input special case) */
duk_uint32_t prev_match_end_coff, prev_match_end_boff;
duk_uint32_t match_start_boff, match_start_coff;
duk_uint32_t match_end_boff, match_end_coff;
DUK_UNREF(thr);
h_input = duk_push_this_coercible_to_string(ctx);
DUK_ASSERT(h_input != NULL);
duk_push_array(ctx);
if (duk_is_undefined(ctx, 1)) {
limit = 0xffffffffUL;
} else {
limit = duk_to_uint32(ctx, 1);
}
if (limit == 0) {
return 1;
}
/* If the separator is a RegExp, make a "clone" of it. The specification
* algorithm calls [[Match]] directly for specific indices; we emulate this
* by tweaking lastIndex and using a "force global" variant of duk_regexp_match()
* which will use global-style matching even when the RegExp itself is non-global.
*/
if (duk_is_undefined(ctx, 0)) {
/* The spec algorithm first does "R = ToString(separator)" before checking
* whether separator is undefined. Since this is side effect free, we can
* skip the ToString() here.
*/
duk_dup_2(ctx);
duk_put_prop_index(ctx, 3, 0);
return 1;
} else if (duk_get_hobject_with_class(ctx, 0, DUK_HOBJECT_CLASS_REGEXP) != NULL) {
#ifdef DUK_USE_REGEXP_SUPPORT
duk_push_hobject_bidx(ctx, DUK_BIDX_REGEXP_CONSTRUCTOR);
duk_dup_0(ctx);
duk_new(ctx, 1); /* [ ... RegExp val ] -> [ ... res ] */
duk_replace(ctx, 0);
/* lastIndex is initialized to zero by new RegExp() */
is_regexp = 1;
#else
DUK_DCERROR_UNSUPPORTED(thr);
#endif
} else {
duk_to_string(ctx, 0);
#ifdef DUK_USE_REGEXP_SUPPORT
is_regexp = 0;
#endif
}
/* stack[0] = separator (string or regexp)
* stack[1] = limit
* stack[2] = input string
* stack[3] = result array
*/
prev_match_end_boff = 0;
prev_match_end_coff = 0;
arr_idx = 0;
matched = 0;
for (;;) {
/*
* The specification uses RegExp [[Match]] to attempt match at specific
* offsets. We don't have such a primitive, so we use an actual RegExp
* and tweak lastIndex. Since the RegExp may be non-global, we use a
* special variant which forces global-like behavior for matching.
*/
DUK_ASSERT_TOP(ctx, 4);
#ifdef DUK_USE_REGEXP_SUPPORT
if (is_regexp) {
duk_dup_0(ctx);
duk_dup_2(ctx);
duk_regexp_match_force_global(thr); /* [ ... regexp input ] -> [ res_obj ] */
if (!duk_is_object(ctx, -1)) {
duk_pop(ctx);
break;
}
matched = 1;
duk_get_prop_stridx(ctx, -1, DUK_STRIDX_INDEX);
DUK_ASSERT(duk_is_number(ctx, -1));
match_start_coff = duk_get_int(ctx, -1);
match_start_boff = duk_heap_strcache_offset_char2byte(thr, h_input, match_start_coff);
duk_pop(ctx);
if (match_start_coff == DUK_HSTRING_GET_CHARLEN(h_input)) {
/* don't allow an empty match at the end of the string */
duk_pop(ctx);
break;
}
duk_get_prop_stridx(ctx, 0, DUK_STRIDX_LAST_INDEX);
DUK_ASSERT(duk_is_number(ctx, -1));
match_end_coff = duk_get_int(ctx, -1);
match_end_boff = duk_heap_strcache_offset_char2byte(thr, h_input, match_end_coff);
duk_pop(ctx);
/* empty match -> bump and continue */
if (prev_match_end_boff == match_end_boff) {
duk_push_int(ctx, match_end_coff + 1);
duk_put_prop_stridx(ctx, 0, DUK_STRIDX_LAST_INDEX);
duk_pop(ctx);
continue;
}
} else {
#else /* DUK_USE_REGEXP_SUPPORT */
{ /* unconditionally */
#endif /* DUK_USE_REGEXP_SUPPORT */
const duk_uint8_t *p_start, *p_end, *p; /* input string scan */
const duk_uint8_t *q_start; /* match string */
duk_size_t q_blen, q_clen;
p_start = DUK_HSTRING_GET_DATA(h_input);
p_end = p_start + DUK_HSTRING_GET_BYTELEN(h_input);
p = p_start + prev_match_end_boff;
h_sep = duk_known_hstring(ctx, 0);
q_start = DUK_HSTRING_GET_DATA(h_sep);
q_blen = (duk_size_t) DUK_HSTRING_GET_BYTELEN(h_sep);
q_clen = (duk_size_t) DUK_HSTRING_GET_CHARLEN(h_sep);
p_end -= q_blen; /* ensure full memcmp() fits in while */
match_start_coff = prev_match_end_coff;
if (q_blen == 0) {
/* Handle empty separator case: it will always match, and always
* triggers the check in step 13.c.iii initially. Note that we
* must skip to either end of string or start of first codepoint,
* skipping over any continuation bytes!
*
* Don't allow an empty string to match at the end of the input.
*/
matched = 1; /* empty separator can always match */
match_start_coff++;
p++;
while (p < p_end) {
if ((p[0] & 0xc0) != 0x80) {
goto found;
}
p++;
}
goto not_found;
}
DUK_ASSERT(q_blen > 0 && q_clen > 0);
while (p <= p_end) {
DUK_ASSERT(p + q_blen <= DUK_HSTRING_GET_DATA(h_input) + DUK_HSTRING_GET_BYTELEN(h_input));
DUK_ASSERT(q_blen > 0); /* no issues with empty memcmp() */
if (DUK_MEMCMP((const void *) p, (const void *) q_start, (size_t) q_blen) == 0) {
/* never an empty match, so step 13.c.iii can't be triggered */
goto found;
}
/* track utf-8 non-continuation bytes */
if ((p[0] & 0xc0) != 0x80) {
match_start_coff++;
}
p++;
}
not_found:
/* not found */
break;
found:
matched = 1;
match_start_boff = (duk_uint32_t) (p - p_start);
match_end_coff = (duk_uint32_t) (match_start_coff + q_clen); /* constrained by string length */
match_end_boff = (duk_uint32_t) (match_start_boff + q_blen); /* ditto */
/* empty match (may happen with empty separator) -> bump and continue */
if (prev_match_end_boff == match_end_boff) {
prev_match_end_boff++;
prev_match_end_coff++;
continue;
}
} /* if (is_regexp) */
/* stack[0] = separator (string or regexp)
* stack[1] = limit
* stack[2] = input string
* stack[3] = result array
* stack[4] = regexp res_obj (if is_regexp)
*/
DUK_DDD(DUK_DDDPRINT("split; match_start b=%ld,c=%ld, match_end b=%ld,c=%ld, prev_end b=%ld,c=%ld",
(long) match_start_boff, (long) match_start_coff,
(long) match_end_boff, (long) match_end_coff,
(long) prev_match_end_boff, (long) prev_match_end_coff));
duk_push_lstring(ctx,
(const char *) (DUK_HSTRING_GET_DATA(h_input) + prev_match_end_boff),
(duk_size_t) (match_start_boff - prev_match_end_boff));
duk_put_prop_index(ctx, 3, arr_idx);
arr_idx++;
if (arr_idx >= limit) {
goto hit_limit;
}
#ifdef DUK_USE_REGEXP_SUPPORT
if (is_regexp) {
duk_size_t i, len;
len = duk_get_length(ctx, 4);
for (i = 1; i < len; i++) {
DUK_ASSERT(i <= DUK_UARRIDX_MAX); /* cannot have >4G captures */
duk_get_prop_index(ctx, 4, (duk_uarridx_t) i);
duk_put_prop_index(ctx, 3, arr_idx);
arr_idx++;
if (arr_idx >= limit) {
goto hit_limit;
}
}
duk_pop(ctx);
/* lastIndex already set up for next match */
} else {
#else /* DUK_USE_REGEXP_SUPPORT */
{ /* unconditionally */
#endif /* DUK_USE_REGEXP_SUPPORT */
/* no action */
}
prev_match_end_boff = match_end_boff;
prev_match_end_coff = match_end_coff;
continue;
} /* for */
/* Combined step 11 (empty string special case) and 14-15. */
DUK_DDD(DUK_DDDPRINT("split trailer; prev_end b=%ld,c=%ld",
(long) prev_match_end_boff, (long) prev_match_end_coff));
if (DUK_HSTRING_GET_CHARLEN(h_input) > 0 || !matched) {
/* Add trailer if:
* a) non-empty input
* b) empty input and no (zero size) match found (step 11)
*/
duk_push_lstring(ctx,
(const char *) DUK_HSTRING_GET_DATA(h_input) + prev_match_end_boff,
(duk_size_t) (DUK_HSTRING_GET_BYTELEN(h_input) - prev_match_end_boff));
duk_put_prop_index(ctx, 3, arr_idx);
/* No arr_idx update or limit check */
}
return 1;
hit_limit:
#ifdef DUK_USE_REGEXP_SUPPORT
if (is_regexp) {
duk_pop(ctx);
}
#endif
return 1;
}
/*
* Various
*/
#ifdef DUK_USE_REGEXP_SUPPORT
DUK_LOCAL void duk__to_regexp_helper(duk_context *ctx, duk_idx_t idx, duk_bool_t force_new) {
duk_hobject *h;
/* Shared helper for match() steps 3-4, search() steps 3-4. */
DUK_ASSERT(idx >= 0);
if (force_new) {
goto do_new;
}
h = duk_get_hobject_with_class(ctx, idx, DUK_HOBJECT_CLASS_REGEXP);
if (!h) {
goto do_new;
}
return;
do_new:
duk_push_hobject_bidx(ctx, DUK_BIDX_REGEXP_CONSTRUCTOR);
duk_dup(ctx, idx);
duk_new(ctx, 1); /* [ ... RegExp val ] -> [ ... res ] */
duk_replace(ctx, idx);
}
#endif /* DUK_USE_REGEXP_SUPPORT */
#ifdef DUK_USE_REGEXP_SUPPORT
DUK_INTERNAL duk_ret_t duk_bi_string_prototype_search(duk_context *ctx) {
duk_hthread *thr = (duk_hthread *) ctx;
/* Easiest way to implement the search required by the specification
* is to do a RegExp test() with lastIndex forced to zero. To avoid
* side effects on the argument, "clone" the RegExp if a RegExp was
* given as input.
*
* The global flag of the RegExp should be ignored; setting lastIndex
* to zero (which happens when "cloning" the RegExp) should have an
* equivalent effect.
*/
DUK_ASSERT_TOP(ctx, 1);
(void) duk_push_this_coercible_to_string(ctx); /* at index 1 */
duk__to_regexp_helper(ctx, 0 /*index*/, 1 /*force_new*/);
/* stack[0] = regexp
* stack[1] = string
*/
/* Avoid using RegExp.prototype methods, as they're writable and
* configurable and may have been changed.
*/
duk_dup_0(ctx);
duk_dup_1(ctx); /* [ ... re_obj input ] */
duk_regexp_match(thr); /* -> [ ... res_obj ] */
if (!duk_is_object(ctx, -1)) {
duk_push_int(ctx, -1);
return 1;
}
duk_get_prop_stridx(ctx, -1, DUK_STRIDX_INDEX);
DUK_ASSERT(duk_is_number(ctx, -1));
return 1;
}
#endif /* DUK_USE_REGEXP_SUPPORT */
#ifdef DUK_USE_REGEXP_SUPPORT
DUK_INTERNAL duk_ret_t duk_bi_string_prototype_match(duk_context *ctx) {
duk_hthread *thr = (duk_hthread *) ctx;
duk_bool_t global;
duk_int_t prev_last_index;
duk_int_t this_index;
duk_int_t arr_idx;
DUK_ASSERT_TOP(ctx, 1);
(void) duk_push_this_coercible_to_string(ctx);
duk__to_regexp_helper(ctx, 0 /*index*/, 0 /*force_new*/);
global = duk_get_prop_stridx_boolean(ctx, 0, DUK_STRIDX_GLOBAL, NULL);
DUK_ASSERT_TOP(ctx, 2);
/* stack[0] = regexp
* stack[1] = string
*/
if (!global) {
duk_regexp_match(thr); /* -> [ res_obj ] */
return 1; /* return 'res_obj' */
}
/* Global case is more complex. */
/* [ regexp string ] */
duk_push_int(ctx, 0);
duk_put_prop_stridx(ctx, 0, DUK_STRIDX_LAST_INDEX);
duk_push_array(ctx);
/* [ regexp string res_arr ] */
prev_last_index = 0;
arr_idx = 0;
for (;;) {
DUK_ASSERT_TOP(ctx, 3);
duk_dup_0(ctx);
duk_dup_1(ctx);
duk_regexp_match(thr); /* -> [ ... regexp string ] -> [ ... res_obj ] */
if (!duk_is_object(ctx, -1)) {
duk_pop(ctx);
break;
}
duk_get_prop_stridx(ctx, 0, DUK_STRIDX_LAST_INDEX);
DUK_ASSERT(duk_is_number(ctx, -1));
this_index = duk_get_int(ctx, -1);
duk_pop(ctx);
if (this_index == prev_last_index) {
this_index++;
duk_push_int(ctx, this_index);
duk_put_prop_stridx(ctx, 0, DUK_STRIDX_LAST_INDEX);
}
prev_last_index = this_index;
duk_get_prop_index(ctx, -1, 0); /* match string */
duk_put_prop_index(ctx, 2, arr_idx);
arr_idx++;
duk_pop(ctx); /* res_obj */
}
if (arr_idx == 0) {
duk_push_null(ctx);
}
return 1; /* return 'res_arr' or 'null' */
}